Students Attention Detection Dataset

Description

The purpose of this dataset is to analyze students’ attention and behavioral patterns during online learning sessions using a comprehensive set of visual and behavioral cues extracted from webcam video streams. The dataset was created through the integration of multiple computer vision modules, including face detection, hand tracking, head pose estimation, gaze tracking, and mobile phone detection. These modules collectively generate high-level behavioral features that characterize student presence, visual focus, body activity, and interaction with potentially distracting objects. The dataset contains 6,244 records and 38 columns, comprising 37 feature columns and one target label column. The target variable (label) indicates whether a student is attentive (0) or inattentive (1). The face-analysis module provides information related to facial presence, face count, face location and dimensions, facial landmark coordinates (eyes, nose tip, and mouth), and face detection confidence. The hand-tracking component extracts the number of detected hands, hand coordinates, and hand–object interaction indicators. The head pose estimation module generates orientation-related attributes, including head pose category, pitch, yaw, and roll angles. The gaze tracking subsystem produces gaze direction, gaze-on-screen status, gaze-point coordinates, and pupil locations for both eyes. The object-detection module identifies mobile phone usage and records phone location and detection confidence. The main features include: • Face Features: face_present, no_of_face, face_x, face_y, face_w, face_h, face_conf • Facial Landmark Features: left_eye_x, left_eye_y, right_eye_x, right_eye_y, nose_tip_x, nose_tip_y, mouth_x, mouth_y • Hand Features: hand_count, left_hand_x, left_hand_y, right_hand_x, right_hand_y, hand_obj_interaction • Head Pose Features: head_pose, head_pitch, head_yaw, head_roll • Mobile Phone Features: phone_present, phone_loc_x, phone_loc_y, phone_conf • Gaze Features: gaze_on_screen, gaze_direction, gazePoint_x, gazePoint_y • Pupil Features: pupil_left_x, pupil_left_y, pupil_right_x, pupil_right_y • Target Label: label Compared with earlier versions, this dataset extends beyond basic face, hand, pose, and phone information by incorporating detailed facial landmarks, gaze estimation, pupil tracking, and hand–object interaction features. These additional behavioral indicators provide richer contextual information about student engagement and visual attention. Consequently, the dataset offers a robust foundation for developing machine learning and explainable AI models for automated attention detection, engagement analytics, formative assessment, and instructor feedback in online higher education environments.

Institutions

• Chittagong University of Engineering and Technology

  Chittagong
• Jahangirnagar University

  Dhaka District, Savar

Categories

Licence